Geological Society of America
Special Paper 404
2006
Monitoring well responses to karst conduit head
fluctuations: Implications for fluid exchange and
matrix transmissivity in the Floridan aquifer
Jennifer M. Martin
Elizabeth J. Screaton
Jonathan B. Martin
Department of Geological Sciences, University of Florida, Box 112120, Gainesville, Florida 32611, USA
ABSTRACT
Karst aquifers with high primary-porosity matrix, such as the Floridan aquifer,
have the potential for movement of water between conduits and matrix, with important
implications for karst development and the maintenance of groundwater quality. The
Santa Fe River Sink and River Rise conduit system, along with the surrounding unconfined Floridan aquifer in north-central Florida, provides a study area to test and quantify conceptual models of exchange between conduits and matrix. The Santa Fe River
sinks underground and flows for ~5 km before reemerging at a first-magnitude spring,
the River Rise. During February and March 2003, we recorded discharge rates into the
Santa Fe River Sink and out of the River Rise along with hydraulic heads at the River
Sink, River Rise, and matrix monitoring wells. Comparison of conduit and monitoringwell hydraulic heads allowed us to track the changes in hydraulic gradient between conduits and wells as a discharge peak passed through the conduits, and the observed head
differences between the wells and conduit show a linear relationship with gains and
losses of water from the conduit system. The responses of heads at three of the monitoring wells to changes in head within the conduits suggest a transmissivity between 950
and 160,000 m2/d, and analysis suggests that the values depend on the scale of measurement. These results demonstrate the potential for transmissivity determinations in karst
aquifers by passive monitoring and are consistent with previous observations that
transmissivity of karst aquifers varies with the scale over which it is measured.
Keywords: karst, aquifer, Floridan, transmissivity.
INTRODUCTION
Hydrologic processes in karst aquifers are clearly important considering that more than a quarter of the world’s population lives on, or obtains its water from, karst aquifers. In the
United States, ~20% of the land surface is karst and 40% of
potable groundwater originates from karst aquifers (Quinlan
and Ewers, 1989). Over the past 40 yr, karst hydrologic
research has undergone a significant shift. From the initial concept that caves were hydrologically isolated from the flow field,
during the 1970s and 1980s, karst hydrology came to signify
primarily the hydrologic properties of conduits (White, 2002).
Martin, J.M., Screaton, E.J., and Martin, J.B., 2006, Monitoring well responses to karst conduit head fluctuations: Implications for fluid exchange and matrix
transmissivity in the Floridan aquifer, in Harmon, R.S., and Wicks, C., eds., Perspectives on karst geomorphology, hydrology, and geochemistry—A tribute volume to Derek C. Ford and William B. White: Geological Society of America Special Paper 404, p. 209–217, doi: 10.1130/2006.2404(17). For permission to
copy, contact [email protected]. ©2006 Geological Society of America. All rights reserved.
209
210
J.M. Martin, E.J. Screaton, and J.B. Martin
In the last decade, however, researchers have begun to realize
that an accurate representation of the hydrologic system must
consider conduit, fracture, and matrix flow components and
describe the relationship among them (White, 2002).
The heterogeneous distribution of porosity and permeability in karst aquifers requires an understanding of the hydrologic
relationship between matrix, fractures, and conduit systems.
This is important when protecting and maintaining groundwater
quality and, in particular, for determining the susceptibility of
karst groundwater resources to surface contaminants. Large
subsurface openings, such as conduits, allow surface water to
travel long distances in a short amount of time with little or no
filtration. When surface runoff containing contaminants flows
only through conduits, karst springs will have high-amplitude,
but relatively brief, periods of water-quality degradation (Smart
and Hobbs, 1986; Ryan and Meiman, 1996). If, on the other
hand, contaminated water infiltrates into the surrounding matrix
porosity, contaminants may have longer residence times.
In aquifers where sinking streams or injections of stormwater runoff into sinkholes cause large fluctuations in head
within the conduit system, exchange of water between conduits,
fractures, and matrix should vary with time. When flow into the
sinkholes is low, conduits are likely to act as a drain for the surrounding fractures and matrix (White, 1999). During times of
high inflow into sinkholes, heads in the conduit system will
rise, forcing water into the surrounding fractures and matrix.
White (1999) speculated that storm response in wells within
and outside of conduits potentially could be used to characterize
coupling between the conduits and the matrix or fractures.
An important control on the exchange of water between
the matrix and conduits is the hydraulic conductivity of the
matrix. The majority of research on karst has been in regions
where extensively recrystallized Paleozoic limestones form the
matrix, resulting in little to no movement of water within the
matrix (White, 1999). In contrast, the relatively young Cenozoic limestones of the Floridan aquifer, and other eogenetic
karst aquifers (Vacher and Mylroie, 2002), have high primary
porosity, producing hydraulic conductivity in the matrix up to
four orders of magnitude greater than in Paleozoic limestones
(Palmer, 2002).
Recession curves of karst spring hydrographs, in which the
discharge declines with time during periods of no excess precipitation, have been used to estimate aquifer properties (e.g.,
Atkinson, 1977). Powers and Shevenell (2000) extended spring
hydrograph analysis methods to study of recession curves for
monitoring wells within karst aquifers. However, analysis of
recession limbs to determine aquifer properties requires knowledge of the distance from the discharge point to the groundwater divide, which may be unknown, or, in the case of a
conduit system linked to a sinking stream, may vary through
time (Powers and Shevenell, 2000). Alternatively, if hydraulic
heads through time are measured within the conduit and nearby
wells, the aquifer parameters can be estimated based on the well
response to changes in head within the conduit.
The Santa Fe River Sink and River Rise conduit system,
and the surrounding unconfined Floridan aquifer, provides a
study area where conceptual models of groundwater exchange
can be tested and quantified. Practical considerations limit us to
distinguishing between large conduit flow, as mapped by cave
divers or inferred from rapid travel times (Screaton et al.,
2004), and matrix flow, which may in reality include intergranular permeability, fractures, and small dissolution features.
In this paper, we present the results from monitoring hydraulic
head fluctuations in the conduits and matrix, and we compare
the head differences between conduit and matrix to calculated
rates of gains or losses of water as it flows through the conduit
system from the River Sink to River Rise. Head and discharge
measurements were also used in combination with analytical
modeling to estimate matrix transmissivity and to describe the
movement of water between the matrix and conduits.
STUDY AREA
The Santa Fe River Basin is a tributary to the Suwannee
River (Fig. 1) and covers an area of ~3583 km2 in north-central
Florida (Hunn and Slack, 1983). Approximately 40 km from its
headwaters at Lake Santa Fe, the Santa Fe River sinks and
flows underground, reappearing intermittently before reemerging ~5 km from the Santa Fe River Sink at a spring called the
Santa Fe River Rise (Fig. 1; Martin and Dean, 2001). This area
of the Santa Fe River is within O’Leno and River Rise Preserve
State Parks, and serves as the field area for the work presented
here (Fig. 1).
Geologic Background
The Floridan aquifer covers an area of ~260,000 km2, and
underlies all of Florida and parts of Georgia, Alabama, and the
southernmost part of South Carolina (Bush and Johnston,
1988). In the Santa Fe River Basin, the Floridan aquifer is composed of Oligocene and Eocene carbonate rocks (Fig. 2) that are
between 90 and 240 m thick (Hunn and Slack, 1983). The
Ocala Limestone is the uppermost unit in the unconfined portion of the Santa Fe River Basin and has an estimated thickness
of ~46–76 m (Hisert, 1994). Surficial sediments of Pleistocene
and Holocene age, composed of white to gray fine sand, ~3 m
thick, cover most bedrock in the Santa Fe Basin. Where present,
the Miocene Hawthorn Group, composed primarily of siliciclastic rocks, acts as a confining unit above the Floridan aquifer.
The erosional edge of the Hawthorn Formation is known as the
Cody Scarp and represents the physical division between the
confined and unconfined Floridan aquifer (Fig. 1). To the northeast of the scarp, where the Hawthorn Group is present, the
Floridan aquifer is confined, and surface water is abundant.
Southwest of the scarp, where the Hawthorn Group has been
removed by erosion, the Floridan aquifer is unconfined or semiconfined, and there are few surface streams. Instead, there are
numerous karst features, such as sinkholes, springs, and disap-
Monitoring well responses to karst conduit head fluctuations
211
Figure 1. Location and schematic map of the study area. (A) Location of Florida (shaded) within the
United States. Box shows location of B. (B) General setting of the Santa Fe River Sink and River
Rise. The Cody Scarp marks the erosional boundary of the Miocene Hawthorn Group. West and
south of the Cody Scarp, the Hawthorn Group has been eroded away, and the Floridan aquifer is
unconfined. Shading north and east of the Cody Scarp indicates the Floridan aquifer is confined.
Dashed lines show generalized potentiometric surface contours based on information for May 2002
from the Suwannee River Water Management District. Contour interval is 5 m. (C) Details of the
study area including the locations of the Santa Fe River Sink and River Rise, monitoring wells, and
mapped and inferred conduits. The shaded area indicates the combined extent of the O’Leno and
River Rise Preserve State Parks.
pearing streams. At the edge of the scarp, streams flowing to the
southwest either flow into sinkholes, as does the Santa Fe
River, or become losing streams. Groundwater flow in the
unconfined Floridan aquifer is generally to the Gulf of Mexico
and the major rivers and springs (Fig. 1B). An extended drought
beginning in 1999 led to very low groundwater levels prior to
this study (Fig. 1B).
Previous Investigations in the Santa Fe River Sink
and River Rise System
The subsurface conduit system between the Santa Fe River
Sink and River Rise has been inferred using natural and introduced tracers, and has also been partially mapped by cave
divers. Hisert (1994) used SF6 as a tracer to connect Santa Fe
River Sink and Sweetwater Lake, and, in a second test, to connect Sweetwater Lake and River Rise (Fig. 1). Dean (1999) and
Martin and Dean (1999) used temperature as a high-resolution
natural tracer and calculated flow rates of 1300–9000 m/d, suggestive of conduits, between the Santa Fe River Sink, Sweetwater Lake, and River Rise. Since 1995, cave divers have
explored conduits within the Santa Fe Sink and Rise system
(Fig. 1). Upstream of Sweetwater Lake, divers have found connections to surface openings located to the east of O’Leno State
Park (Fig. 1). Cave divers have estimated the average dimensions of the passages upstream of River Rise to range in width
from 18 to 24 m and in height from 12 to 18 m. Depths of the
flow of the conduit typically range from 30 to 40 m below the
water table (Poucher, 2001).
In a recent study, Screaton et al. (2004) used temperature
signals to connect flow and determine velocities between additional surface-water features, not yet connected by cave divers.
Connecting these surface-water features helped to infer a conduit path between the Santa Fe River Sink and Sweetwater
212
J.M. Martin, E.J. Screaton, and J.B. Martin
of water levels. Using the best-fit curve of the data points, it
was possible to infer discharges for all water levels within the
range of measured values.
Monitoring Wells
Figure 2. Lithostratigraphic and hydrostratigraphic units of the Santa
Fe Basin based on Hunn and Slack (1983).
During early 2003, seven monitoring wells were installed
at various distances from mapped or inferred subsurface conduits (Fig. 1). The monitoring wells were constructed of 2-in
(0.05 m)-diameter PVC set in a 6-in (0.15 m)-diameter borehole and were ~30 m deep each. Each well had a 20 ft (6.1 m)
screened interval. The annular space around the screened interval was filled with sand up to ~1 m above the screened interval.
The sand pack was topped with bentonite, and the remaining
annular space was filled with grout. Well development was performed by surging the well and pumping out water to remove
fines. An existing well, Tower Well, located near the main
entrance to O’Leno State Park, was also monitored. The total
depth of Tower Well is 26 m, and the casing depth is not known.
Water Levels
Lake (Fig. 1). In addition, Screaton et al. (2004) compared estimated inflow rates of the Santa Fe River into the River Sink
with outflow rates from the conduit system, as determined from
the discharge of the Santa Fe River immediately downstream of
River Rise. Results suggest that the conduit system gains water
during times when input into the Santa Fe River Sink is low,
and loses water during times of high flow into the Santa Fe
River Sink. These results are consistent with previous studies
of water chemistry at the River Sink and Rise, which indicated
groundwater contributions from the matrix to the conduit system
under low-flow conditions (Skirvin, 1962; Martin and Dean,
2001), and chemical changes at a well near the conduit system,
which suggested loss of conduit water to the matrix following a
flood event (Martin and Dean, 2001). However, Screaton et al.
(2004) could not evaluate how gains or losses from the conduits
were related to flow out of or into the aquifer matrix because of
a lack of monitoring wells outside of the conduits during their
investigation.
METHODS
Santa Fe River Sink and River Rise Discharge
Discharge rates of the Santa Fe River entering the River
Sink were calculated by converting water levels to discharge
using a rating curve developed by the Suwannee River Water
Management District (SRWMD, Rating No. 3 for Station Number 02321898, Santa Fe River at O’Leno State Park). River Rise
discharge rates were calculated using a relationship between
water-level elevations and unpublished discharge data from
SRWMD. The rating curve was constructed by Screaton et al.
(2004) by plotting recorded discharge measurements for a variety
Water levels were monitored at the Santa Fe River Sink and
River Rise to represent conduit hydraulic heads. Water levels at
monitoring wells were assumed to represent hydraulic head in the
matrix. Observations during drilling of wells 1 through 7 did not
indicate any penetration of large conduits (e.g., there were no significant bit drops). Tower Well was assumed to represent matrix
due to its significant distance from any mapped or inferred conduits. As noted already, our categorization of matrix may include
thin conduits and fractures as well as intergranular porosity.
Water levels were monitored by one of three types of
recording instruments (Global Water WL14 Water Level Logger
with an accuracy of ±0.01 m, Van Essen Diver with an accuracy
of ±0.005 m, or Van Essen CTD Diver with an accuracy of
±0.03 m) or were measured using an electronic probe. The
water-level loggers located at the River Sink and River Rise
were installed within 2-in (0.05 m)-diameter, slotted PVC pipes
fixed to trees on the shore. The water-level loggers at monitoring wells were attached to the well cap by a plastic-coated stainless steel wire. Water levels were generally recorded at 10 min
intervals. Data were downloaded from the loggers every four to
five weeks. For each recording interval, water pressures from
the Van Essen Divers and CTD Divers were corrected using the
ambient barometric pressure recorded by a Van Essen Baro
Diver© (with an accuracy of ±0.0045 m) and then referenced
to the water elevation surveyed at the time the data were downloaded or measured in the wells.
Reference elevations at the Santa Fe River Sink and River
Rise were surveyed by personnel from the Department of Geological Sciences, University of Florida, using a Sokkia Automatic Level Model B21. These surveys were tied to nearby U.S.
Army Corp of Engineers bench marks. Temporary bench marks
(nails in trees) were established at each monitoring location,
Monitoring well responses to karst conduit head fluctuations
and water levels were measured monthly relative to the temporary bench marks. The monitoring wells were surveyed by commercial surveyors.
Water-level discrepancies at the Santa Fe River Sink and
River Rise were examined by comparing consistency of
recorded values with manual water-level readings taken at the
beginning and end of a recording interval. Observed discrepancies were typically <±0.03 m, and likely reflect movement of
the logger or instrument drift. Total estimated errors were calculated by summing the observed discrepancies, instrument
error, and survey errors, and were estimated to be <±0.07 m and
<±0.08 m for the River Sink and River Rise, respectively.
Water-level errors at the monitoring wells are expected to be
lower than at the karst windows because pressure transducer
movement, a suspected source of error at the surface-water sites
due to human or natural disturbance of the PVC, is likely to be
minimal within the monitoring wells. Summation of survey
error, water-level reading error, and instrument error suggests
total errors at the monitoring wells of ±0.02 m for manual readings and ±0.03 m for automated readings.
The method of Pinder et al. (1969) assumes that the aquifer
is homogeneous and of uniform thickness, and that the stream
fully penetrates the entire thickness of the aquifer. It should be
noted that these assumptions are not strictly met in this analysis.
The aquifer is probably heterogeneous, and the conduits and
monitoring wells do not penetrate the full thickness of the Floridan aquifer. Estimates of conduit dimensions by cave divers
suggest heights ranging from 12 to 18 m, while the total thickness of the Floridan aquifer is estimated to be 90–240 m. Thus,
vertical flow effects are likely, especially for wells close to the
conduit. The analysis also does not account for head changes at
the wells due to diffuse recharge. However, the methods used in
this study provide a first approximation of the hydrologic properties of the system.
Because the Floridan aquifer system in the Santa Fe River
Basin is not known to be laterally bounded by impermeable
materials, an equation for a semi-infinite aquifer was used
(Pinder et al., 1969). The input signal, which is the conduit
hydraulic head, is broken into increments, and then the incremental change in head is calculated:
⎛ x ⎞
,
∆hm = ∆H m erfc ⎜
⎜ 2 tT S ⎟⎟
⎝
⎠
Matrix Transmissivity
Accurate estimates of transmissivity are necessary to predict
aquifer response to various hydrologic stresses (Pinder et al.,
1969) and can be used to estimate hydraulic conductivity by
dividing transmissivity by aquifer thickness. Transmissivity is
difficult to determine through pumping tests in karst aquifers
due to the very high rates of pumping necessary to create a significant drawdown of the potentiometric surface during aquifer
tests. Passive monitoring uses naturally occurring events to
determine aquifer properties, and can provide an inexpensive
and more representative alternative to pumping tests (Powers
and Shevenell, 2000).
Methods have been developed to determine aquifer properties using response to tidal fluctuations (e.g., Ferris, 1963). It
has been suggested that these methods could be applied to a single event, such as a storm pulse in a stream (Ferris, 1963).
Cooper and Rorabaugh (1963) derived solutions for the change
in groundwater head to changes in river stage, approximated by
a sinusoidal hydrograph. Pinder et al. (1969) developed a
method that expanded analysis of the transmissivity of aquifers
surrounding streams to allow use of non-sinusoidal hydrographs. In this study, we apply the method of Pinder et al.
(1969) to determine aquifer properties surrounding the Santa Fe
River Sink and River Rise conduit system. One difficulty with
applying this method to a conduit system is that the exact locations of the conduits are often not known. For the Santa Fe
River Sink and River Rise system, conduit pathways have been
mapped by cave divers (Fig. 1) and inferred from previous studies of temperature transmission between the River Sink, karst
windows, and the River Rise (Martin and Dean, 2001; Screaton
et al., 2004). However, additional conduits that have not yet
been mapped are likely to exist throughout the region.
213
(1)
where ∆hm is the incremental change in head at the well in the
mth time step, ∆Hm is the incremental change in head at the conduit in the mth time step, x is the distance between the well and
the conduit (m), T is the transmissivity (m2/d), S is the storativity
(dimensionless), erfc is the complementary error function, and t
is the time-step size (d). The total change in head at the well at
any time is calculated by summing the increments of head
change resulting from all prior changes in the input signal.
For the unconfined Floridan aquifer, the appropriate storage
parameter (S) would be specific yield. A value of 0.20 was
chosen as a reasonable estimate of the storage parameter, based
on reported porosity values ranging from 0.1 to 0.45 (Palmer,
2002). Distances from the wells to the conduits were estimated
based on the closest mapped or inferred conduit location, and
thus should be considered a maximum value.
RESULTS
Santa Fe River Sink and River Rise Discharge
Plotting the flow into the Santa Fe River Sink and the outflow from River Rise (Fig. 3) illustrates the gains or losses from
this reach of the conduit system. Prior to 17 November 2002,
both the inflow to the Santa Fe River Sink and the outflow from
River Rise were negligible due to drought conditions. During
most of the study period in which there was flow, the discharge
out of Santa Fe River Rise exceeded the flow rate into Santa Fe
River Sink. In contrast, the inflow into the Santa Fe River Sink
exceeded the discharge out of River Rise during the two time
intervals marked by shading on Figure 3.
214
J.M. Martin, E.J. Screaton, and J.B. Martin
during this period, total rainfall recorded at O’Leno State Park
was only 0.27 m (Fig. 3). The change in head at the wells
appears to be transmitted from the conduit system. Furthermore, water-level maxima in the wells lag the water-level maximum in the Santa Fe River Sink. The time lag is a reflection of
the distance, transmissivity, and specific yield between the conduit and each well and was used to calculate transmissivity of
the Floridan aquifer matrix.
Transmissivity
Figure 3. Precipitation and Santa Fe River discharge data for November 2002 to March 2003. (A) Daily precipitation recorded at O’Leno
State Park. (B) Inflow rates into Santa Fe River Sink and discharge
rates from River Rise. Shaded areas show time intervals when output
from the River Rise exceeded input to the River Sink.
Observed hydrographs at Tower Well, well 1, and well 4
were compared to changes in head calculated using the method
of Pinder et al. (1969). Transmissivity was adjusted until the
calculated curves matched the measured hydrographs (Fig. 5).
Matching the curves was done visually. Curve matching could
not be performed for well 2 due to lack of data during the peak
Matrix Water Levels
Water levels in monitoring wells and at Santa Fe River
Sink are shown in Figure 4. During the monitoring period, there
was high inflow into Santa Fe River Sink (Fig. 3), resulting in
high water levels, which we compare with the matrix water
levels from the monitoring wells (Fig. 4). Water-level fluctuations in the matrix were too large to be explained by diffuse
recharge alone. For example, the water level at well 4 fluctuated ~2 m between 8 February 2003 and 17 March 2003, but
Figure 4. Water levels measured at the River Sink and monitoring
wells January to March 2003. Continuous records (every 10 min) were
available from Tower Well and well 4. Water levels are reported in m
above sea level (asl; 1927 North American Datum). The data record
from well 1 is continuous except for missing data between 13 February
and 5 March 2003, while the record from well 2 is missing data after
26 February 2003. Wells 3 and 6 have only intermittent probe
measurements, and data from well 7 were insufficient to plot. Well 5
installation was not completed until after the storm event.
Figure 5. Comparison of measured and calculated water levels for (A)
Tower Well, (B) well 4, and (C) well 1. T—transmissivity.
Monitoring well responses to karst conduit head fluctuations
of the storm response. Matching of limited data at wells 3 and 6
suggests that these wells were also responding to conduit head
changes (Martin, 2003), but data were too limited to confidently
assess transmissivity.
The conduit head during a time step (Hm) was assumed
equal to the measured water level at the Santa Fe River Sink for
matching of Tower Well and well 1 hydrographs, and to the
measured water levels of River Rise for analysis of the well 4
hydrograph. It was assumed that the magnitude of head changes
after 6 February 2003 (up to 0.63 m/d) was large enough to disregard any effects of head changes prior to the start of the
analysis (<0.01 m/d). Loss of River Sink and River Rise data
prevented continuation of the analysis beyond 27 March 2003.
Calculated transmissivity ranged from 950 to 160,000 m2/d
(Table 1). Equation 1 indicates that if higher or lower values
of specific yield were used for the curve matching, the optimal
transmissivity would change proportionally. Thus, transmissivity may be ~0.5–2 times the calculated values, if specific
yield were as low as 0.10 or as high as 0.40. If the actual distance to the nearest conduit is less than the value used to calculate transmissivity, the transmissivity will decrease. For
example, if the distance to the nearest conduit were half of the
estimated value, optimal transmissivity would be four times
smaller than estimated.
DISCUSSION
Transmissivity Values
Averaging numerous small-scale tests of transmissivity in a
karst aquifer yields lower values than the results of large-scale
tests in the same area (Bradbury and Muldoon, 1990; Rovey
and Cherkauer, 1995). This scaling effect stems from the likelihood that large-scale tests will include preferential paths
through the matrix, and these preferential pathways will dominate a larger percentage of groundwater flow, thus increasing
average transmissivity (Rovey, 1994). In karstic carbonates,
such as the Floridan aquifer, transmissivity increases with the
amount of dissolution within the aquifer (Rovey, 1994). In addition to the possible effects of the partial penetration of the conduit, scale effects are a likely reason that the transmissivity
calculated for well 4 (a distance of 115 m from the conduit) is
two orders of magnitude smaller than transmissivity calculated
at well 1 and Tower Well, with distances of 475 and 3750 m
from the conduit, respectively. At a smaller scale, slug tests performed on wells 3, 4, 5, 6, and 7 yielded transmissivity ranging
from 270 m2/d to 550 m2/d (Hamilton, 2003, personal commun.),
which is lower than that calculated from the storm response for
215
well 4. At a regional scale, Bush and Johnston (1988) estimated
a transmissivity of 93,000 m2/d for the area containing the
Santa Fe River Sink and River Rise System, based on calibration of a finite-difference model with a cell size of 165 km2.
This large-scale result is similar in magnitude to transmissivity
calculated for well 1 and Tower Well (Table 1) using the
method of Pinder et al. (1969).
Mixing of Conduit and Matrix Water
Differences between the rate of water entering the Santa Fe
River Sink and emerging from the River Rise indicate that
water enters or leaves the conduit system along the reach
between the River Sink and River Rise (Fig. 3). Although there
is a conduit entering the main conduit from the east (Fig. 1) and
probably other unmapped conduits, there are no obvious surface-water sources supplying water to these systems. Thus, at
low-flow conditions, the conduit system appears be recharged
by groundwater from the surrounding matrix. Previous observations of the chemical composition and temperature of River
Rise water are consistent with this interpretation (Martin and
Dean, 2001). In contrast, when inflow into the Santa Fe River
Sink exceeds the outflow from River Rise, the conduit system
must be losing water, possibly into the matrix porosity of the
surrounding aquifer.
To further explore the relationship between the conduit
system and the surrounding aquifer, the difference between
inflow into the Santa Fe River Sink and discharge out of River
Rise was plotted versus the corresponding head difference
between a well and the nearest conduit measuring point
(Fig. 6). Tower Well, well 4, and well 1 were analyzed
because of the high number of data points collected from each
location. Well 2, estimated to be ~1000 m from the conduit
system, had less data, but showed similar results. The analyses
indicate that when the conduit system gains water, the head
measured in the observation wells is higher than within the
conduit system. Conversely, when the conduit system loses
water, the head measured in the monitoring wells is lower than
within the conduit system. Ideally, the best-fit line should
cross the origin, which is the point where the conduit is neither gaining nor losing water and the head difference is zero.
Summation of the errors of the water-level measurements at
the Santa Fe River Sink or River Rise and the wells suggests a
total error of ±0.09–0.11 m. Results for wells 1 and 2 appear
within the range of errors. Well 4 head is 0.21 m greater than
the conduit (Fig. 6B) when there is no net conduit gain or loss,
which is slightly larger than the estimated error. Tower Well
(Fig. 6A) head is 0.38 less than the conduit when there is no
net conduit gain or loss, which is greater than could be attributed to water-level measurement error. One possibility for this
discrepancy is that the conduit may not be gaining or losing
water uniformly along its length. This explanation could be
tested through measurements made at a smaller scale than the
current study.
216
J.M. Martin, E.J. Screaton, and J.B. Martin
Figure 6. Relationship between change in conduit gains and losses and head differences for (A) Tower Well, (B) well 4, (C) well 1, and (D)
well 2.
Although the data show some scatter, the plots also indicate a linear relationship between head differences and the net
gain or loss of water from the conduit system for all the wells
(Fig. 6), suggesting that flow may be Darcian between the conduits and wells. Further interpretation of the slopes in Figure 6
is limited by not knowing the portion of the total net gain or
loss between Santa Fe River Sink and River Rise that is represented by the measurements at each well. Due to heterogeneity,
a greater percentage of the net flow into or out of the conduit
would be expected in areas of high transmissivity and less
inflow or outflow in areas of low transmissivity.
SUMMARY
Karst aquifers are a significant source of drinking water for
millions of people, but because of their high permeability and
connection to surface-water sources, they are especially vulnerable to contamination by surface water. Understanding the relationship between the exchange of matrix and conduit water will
help in determining the best way to protect this valuable
resource. Surface water entering the matrix will also have significant effects on karstification within the Floridan aquifer
because of the differences in the chemical composition of the
surface and groundwater.
Floridan aquifer transmissivity over several distances
between conduits and monitoring wells was quantified using the
passive monitoring method of Pinder et al. (1969). Calculated
values are between 950 and 160,000 m2/d. The variability in
transmissivity with distance from the conduit is consistent with
previous observations (Bradbury and Muldoon, 1990; Rovey
and Cherkauer, 1995; Rovey, 1994), which showed that measured transmissivity in karst regions increases with increasing
scale. Comparison of conduit and monitoring-well hydraulic
head allowed tracking of the changes in hydraulic gradient as a
storm peak passed through the conduit. Our results confirm the
suggestion by White (1999) that storm response in wells within
and outside of conduits could potentially be used to characterize
coupling between conduits and matrix or fractures.
Monitoring well responses to karst conduit head fluctuations
ACKNOWLEDGMENTS
Acknowledgment is given to the Florida Department of
Environmental Protection and the staff of O’Leno State Park for
their cooperation. Reviews by Kevin Cunningham and Andy
Long improved this manuscript. The research was funded by
the National Science Foundation grants EAR-003360 and
EAR-0510054.
REFERENCES CITED
Atkinson, T.C., 1977, Diffuse and conduit flow in limestone terrain in the
Mendip Hills, Somerset (Great Britain): Journal of Hydrology, v. 35,
p. 93–110, doi: 10.1016/0022-1694(77)90079-8.
Bradbury, K.R., and Muldoon, M.A., 1990, Hydraulic conductivity determinations in unlithified glacial and fluvial materials, in Nielson, D.M., and
Johnson, A.I., eds., Hydraulic conductivity and waste contaminant
transport in soils: Philadelphia, American Society for Testing and Materials, v. 1142, p. 138–151.
Bush, P.W., and Johnston, R.H., 1988, Ground-water hydraulics, regional flow,
and ground-water development of the Floridan aquifer system in Florida
and in parts of Georgia, South Carolina, and Alabama: U.S. Geological
Survey Professional Paper 1403-C, 80 p.
Cooper, H.H., and Rorabaugh, M.J., 1963, Groundwater movements and bank
storage due to flood stages in surface streams: U.S. Geological Survey
Water-Supply Paper 1536-J, 23 p.
Dean, R.W., 1999, Surface and groundwater mixing in a karst aquifer: An
example from the Floridan aquifer [M.S. thesis]: Gainesville, University of Florida, 74 p.
Ferris, J.G., 1963, Cyclic water-level fluctuations as a basis for determining
aquifer transmissibility, in Bentall, R., ed., Methods of determining permeability, transmissibility and drawdown: U.S. Geological Survey
Water-Supply Paper 1536-I, p. 305–323.
Hisert, R.A., 1994, A multiple tracer approach to determine the ground and surface water relationships in the western Santa Fe River, Columbia
County, Florida [Ph.D. thesis]: Gainesville, University of Florida, 212 p.
Hunn, J.D., and Slack, L.J., 1983, Water resources of the Santa Fe River Basin,
Florida: U.S. Geological Survey Water-Resources Investigations Report
83-4075, 105 p.
Martin, J.B., and Dean, R.W., 1999, Temperature as a natural tracer of short
residence times for ground water in karst aquifer, in Palmer, A.N.,
Palmer, M.V., and Sasowsky, I.D., eds., Karst modeling: Charlestown,
West Virginia, Karst Waters Institute, p. 236–242.
Martin, J.B., and Dean, R.W., 2001, Exchange of water between conduits and
matrix in the Floridan aquifer: Chemical Geology, v. 179, p. 145–165,
doi: 10.1016/S0009-2541(01)00320-5.
Martin, J.M., 2003, Quantification of matrix hydraulic conductivity in the Santa
Fe River Sink/Rise system with implications for the exchange of water
217
between the matrix and conduits [M.S. thesis]: Gainesville, University of
Florida, 80 p. Online at http://etd.fcla.edu/UF/UFE0002882/martin_j.pdf.
Palmer, A.N., 2002, Karst in Paleozoic rocks: How does it differ from Florida?,
in Martin, J.B., Wicks, C.M., and Sasowsky, I.D., eds., Hydrogeology
and biology of post-Paleozoic carbonate aquifers: Charlestown, West
Virginia, Karst Water Institute, p. 185–191.
Pinder, G.F., Bredehoeft, J.D., and Cooper, H.H., Jr., 1969, Determination of
aquifer diffusivity from aquifer response to fluctuations in river stage:
Water Resources Research, v. 5, p. 850–855.
Poucher, M., 2001, Cave exploration at River Rise and Sweetwater Lake, project
update September 1, 2001: http://cavesurvey.com/sept_1,_2000.htm
(accessed 16 January 2003).
Powers, J.G., and Shevenell, L., 2000, Transmissivity estimates from well hydrographs in karst and fractured aquifers: Ground Water, v. 38, p. 361–369,
doi: 10.1111/j.1745-6584.2000.tb00221.x.
Quinlan, J.F., and Ewers, R.O., 1989, Subsurface drainage in the Mammoth
Cave area, in White, W.B., and White, E.L., eds., Karst hydrology: Concepts from the Mammoth Cave area: New York, Van Nostrand Reinhold, p. 65–103.
Rovey, C.W., II, 1994, Assessing flow systems in carbonate aquifers using
scale effects in hydraulic conductivity: Environmental Geology, v. 24,
p. 244–253, doi: 10.1007/BF00767085.
Rovey, C.W., II, and Cherkauer, D.S., 1995, Scale dependency of hydraulic
conductivity measurements: Ground Water, v. 33, p. 769–780, doi:
10.1111/j.1745-6584.1995.tb00023.x.
Ryan, M., and Meiman, J., 1996, An examination of short-term variations in
water quality at a karst spring in Kentucky: Ground Water, v. 34, p. 23–30,
doi: 10.1111/j.1745-6584.1996.tb01861.x.
Screaton, E., Martin, J.B., Ginn, B., and Smith, L., 2004, Conduit properties
and karstification in the Santa Fe River Sink-Rise system of the Floridan
aquifer: Ground Water, v. 42, p. 338–346, doi: 10.1111/j.1745-6584.
2004.tb02682.x.
Skirvin, R.T., 1962, The underground course of the Santa Fe River near High
Springs, Florida [M.S. thesis]: Gainesville, University of Florida, 52 p.
Smart, P.L., and Hobbs, S.L., 1986. Characterization of carbonate aquifers: A
conceptual base, in Environmental Problems in Karst Terranes and
Their Solutions Conference, Bowling Green, Kentucky, United States,
October 28–30, 1986: Dublin, Ohio, National Water Well Association,
p. 1–14.
Vacher, H.L., and Mylroie, J.E., 2002, Eogenetic karst from the perspective of an
equivalent porous medium: Carbonates and Evaporites, v. 17, p. 182–196.
White, W.B., 1999, Conceptual models for karstic aquifers, in Palmer, A.N.,
Palmer, M.V., and Sasowsky, I.D., eds., Karst modeling: Charlestown,
West Virginia, Karst Waters Institute, p. 11–16.
White, W.B., 2002, Karst hydrology: Recent developments and open questions: Engineering Geology, v. 65, p. 85–105, doi: 10.1016/S0013-7952
(01)00116-8.
MANUSCRIPT ACCEPTED BY THE SOCIETY 22 SEPTEMBER 2005
Printed in the USA